Determining a safety configuration for a robot workstation

Abstract

A method (100) for determining a safety configuration (2) for a robot workstation (1) that comprises at least one industrial robot (3), wherein the safety configuration (2) comprises a combination of at least one monitored safety zone (2a) and at least one velocity limit (7a-7f) of the robot (3) that are arranged such that, at the moment where reaching or other entry of a person into the monitored safety zone (2a) is detected, there is sufficient time to stop the robot (3) before it can possibly come into contact with the entering person, the method (100) comprising the steps of: - determining (110) locations (4a) in space, as well as velocities (4b), of a least one reference point (4) of the robot (3) at multiple trajectory points (6) along a present or future trajectory (5) of the robot (3); - dividing (120) the trajectory points (6) into groups (6a-6f); - for each group (6a-6f), determining (130) a velocity limit (7a-7f) based at least in part on the maximum velocity of the reference point at any of the trajectory points (6) in this group (6a-6f); and - determining (140) the safety configuration (2) based at least in part on the locations (4a) of the reference point (4) of the robot (3) at the trajectory points (6) on the one hand, and the velocity limits (7a-7f) applicable to these trajectory points (6) on the other hand.

Description

[0001] ABB Schweiz AG 22.08.2024

[0002] A 19252 WO

[0003] DETERMINING A SAFETY CONFIGURATION FOR A ROBOT WORKSTATION

[0004] FIELD OF THE INVENTION

[0005] The invention relates to the operation of robots on a factory floor where there is a need to protect human workers from injury by the moving robot.

[0006] BACKGROUND

[0007] Industrial robots move at great speeds and with great force to accomplish their assigned task. A collision with a person on the factory floor can cause severe injury or even death. Therefore, the robot workstation has to be safeguarded against such collisions. A straight-forward way is to fence in the whole area that any part of the robot is theoretically able to reach. But this uses very much space on the factory floor and also obstructs useful interactions of humans with the robot, such as providing materials to the robot and taking away products assembled by the robot.

[0008] Therefore, many safety concepts for robot workstations rely on creating a monitored safety zone around the robot. Entry into the safety zone is detected by a light curtain, a laser scanner or other means. In response to such a detection, the robot is slowed down or stopped. The size of the safety zone needs to be dimensioned such that the robot can stop in time before reaching the entering person. It thus depends on the moving speed of the robot. In this manner, collaborative operations involving robots and persons are facilitated, and less space on the factory floor is used. This speed and separation monitoring is specified in ISO / TS 15066.

[0009] OBJECTIVE OF THE INVENTION

[0010] It is an objective of the invention to further improve safety configurations for robot workstations while at the same time facilitating the automated creation of such safety configurations. P240530W001 - 2 - 22.08.2024

[0011] This objective is achieved by a method according to the independent claim. Further advantageous embodiments are detailed in the dependent claims.

[0012] DISCLOSURE OF THE INVENTION

[0013] The invention provides a method for determining a safety configuration for a robot workstation. This robot workstation comprises at least one industrial robot. The safety configuration comprises a combination of at least one monitored safety zone and at least one velocity limit of the robot that are arranged such that, at the moment where reaching or other entry of a person into the monitored safety zone is detected, there is sufficient time to stop the robot before it can possibly come into contact with the entering person. That is, the faster the robot is moving, the larger the stopping distance is, and the larger the monitored safety zone needs to be.

[0014] In the course of the method, locations in space, as well as velocities, of a least one reference point of the robot at multiple trajectory points along a present or future trajectory of the robot are determined. For example, the reference point of the robot may be a tool center point, TCP, or a wrist or elbow of the robot. The locations and velocities may, for example, be obtained by simulating a trajectory of the at least one reference point of the robot. This is particularly useful to assess a multitude of motion scenarios for a “what if... ” evaluation. But the locations and velocities may just as well be measured, e.g., in a test run of the robot.

[0015] The trajectory points are divided into groups. For each group, a velocity limit is determined based at least in part on the maximum velocity of the reference point at any of the trajectory points in this group. For example, this maximum velocity may be set as the velocity limit, or the velocity limit may be set to said maximum velocity minus an additional safety margin.

[0016] The sought safety configuration is then determined based at least in part on the locations of the reference point of the robot at the trajectory points on the one hand, and the velocity limits applicable to these trajectory points on the other hand.

[0017] The inventors have found that the dividing of the trajectory points into groups allows to differentiate the safety measures between different parts of the trajectory. In many robot applications, very stringent measures may need to be applied only along a small P240530W001 - 3 - 22.08.2024 part of the trajectory, whereas more lenient measures may suffice along the rest of the trajectory.

[0018] For example, when a robot moves across its cell on the factory floor to a place where it is to accomplish a task, the motion will initially be slow, then reach a maximum on the coast towards the intended place, and then slow down again when it reaches its intended position. The part of the trajectory where a larger separation distance is required is the part where the robot moves fastest. But at the intended position where the work is being performed, the robot moves much slower, so persons may be allowed to come closer to it, and a more collaborative work experience may be allowed.

[0019] Likewise, where the available separation distance between the intended robot trajectory and any persons is fixed and to be taken as a given, there may be only certain areas along the trajectory where the available space is particularly tight, whereas there may be much more space available along the rest of the trajectory. It may then be necessary to slow down the robot for the sake of safety only where the available space is tight. In this manner, the productivity of the robot in terms of “units useful work per unit time” is improved.

[0020] In theory, the mentioned advantages could be cranked up to the maximum by making the groups as small as possible, namely comprising only one single trajectory point. But there is a practical limit to this due to the important role that the velocity limits play. In a particularly advantageous embodiment, zones in the space of the trajectory points corresponding to the groups, as well as velocity limits corresponding to these groups, are loaded into a safety controller of the robot. This safety controller is configured to enforce these velocity limits independently of a motion controller of the robot. That is, whenever the trajectory enters a particular zone, the robot is forced to stay below the respective velocity limit. This is somewhat akin to a functionality of the German PZB train safety system: When the train is approaching a pre-signal indicating that the main signal 1000-1500 meters ahead will be red, in the remaining distance to the main signal, there are multiple check points at which the speed of the train must comply with progressively lower limits. Akin to the PZB system, the enforcing of the velocity limits by the safety controller is very important. Therefore, the enforcing of each and every velocity limit must be validated. If the trajectory comprised, e.g., 100 individual trajectory points, and each trajectory point was associated with its own velocity limit, P240530W001 - 4 - 22.08.2024 then validation measurements would have to be procured for each and every such velocity limit, and a validation would have to be performed accordingly. This would cost more time than could, e.g., be gained by letting the robot run still a little bit faster because of the maximally fine-grained groups.

[0021] Furthermore, if the groups comprised only one single trajectory point each, then spatial zones in the space of the trajectory points corresponding to the groups might be smaller than the spatial dimensions of the robot. The monitoring, by the safety controller, where the robot is would then yield the result that the robot is in multiple zones at once. One would therefore, e.g., run into difficulties as to which of multiple velocity limits applicable in these multiple zones should be applied to the motion of the robot.

[0022] Thus, there is an optimal group size in terms of numbers of trajectory points. Corresponding to this optimal group size, there is an optimal size of zones in the space of the trajectory points.

[0023] The possibility to simulate “what-if...” scenarios according to the present method helps a great deal to automate the determining of the safety configuration. Previously, the manual configuration of the safety zones and the velocity limits was a challenging process. Setting the limits for the supervised velocity much higher than the achievable speed of the robot during the programmed trajectory results in longer stop durations and thus in an unnecessary large safety sensor monitoring area. Setting the velocity limits very low decreases productivity. For a given trajectory, finding safety zones and velocity limits that minimize the sensor monitoring area (and therefore the cell footprint) without reducing the robot speeds is an iterative and time-consuming process. In addition, a change in the trajectory would require new iterations of this process to adapt the safety zones and velocity limits. A similar challenge occur when a sensor monitoring area is given, either to avoid reprogramming the safety sensors or to satisfy a specific cell footprint. Finding an efficient configuration of safety zones and velocity limits for a given sensor monitoring area while satisfying ISO / TS 15066 is a timeconsuming process for the user.

[0024] In a particularly advantageous embodiment, the determining of the safety configuration comprises determining, for each trajectory point, based on the velocity limit applicable P240530W001 - 5 - 22.08.2024 to this trajectory point, a required separation distance for persons from the location of the reference point of the robot at this trajectory point. As discussed before, this required separation distance may, for example, be motivated by the stopping distance that is required for stopping motion of the robot moving at the velocity limit from the trajectory point on. For calculating this stopping distance, in most present applications, the shortest distance between a human and the robot prevails, whereas directions of motion are not taken into account. That is, even if the robot is presently moving away from the human, the required separation distance is still the same.

[0025] In particular, as part of the determined safety configuration, at least one monitored safety zone may be determined based at least in part on the required separation distances. For example, circles (or spheres) with a radius corresponding to the respective required separation distance may be drawn around each trajectory point, and the monitored safety zone may be chosen to comprise at least the union of these circles or spheres. For example, this monitored safety zone may be an envelope of this union, or a geometric shape may be chosen that is slightly larger than this envelope (meaning that the envelope fully fits within it) but simpler to describe, such as a bounding box. Instead of circles or spheres, other suitable forms, such as convex hulls, may be used just as well.

[0026] In a further particularly advantageous embodiment, in the course of determining the safety configuration based at least in part on the required separation distances, the required separation distances, and / or an aggregate thereof, is compared with at least one given monitored safety zone. In response to determining that a required separation distance for a trajectory point exceeds the monitored safety zone, the velocity limit for the group to which this trajectory point belongs is reduced. In this manner, the required separation distance is reduced as well and may therefore be brought back to within the confines of the given monitored safety zone. In this manner, the safety configuration may be optimized such that the robot can move and perform its work as fast as possible. For example, an initial safety configuration may start out with the maximum desirable velocity limits, and then these velocity limits may be reduced only where exactly this is needed to ensure safety of persons.

[0027] In this context, reducing the velocity limit for a particular group may also comprise dividing this group into several groups and setting different velocity limits for these P240530W001 - 6 - 22.08.2024 several groups. In this manner, the total reduction in velocity, and thus the total increase in cycle time along the trajectory of the robot, may be kept to a minimum.

[0028] In a particularly advantageous embodiment, at least one group of trajectory points is formed by unsupervised clustering. As discussed before, in typical robot applications, motion of the robot will occur in distinct phases, such as accelerating, coasting towards an intended work area, and decelerating when reaching the work area. These phases may be automatically recognized by the unsupervised clustering. For example, the clustering may be performed by k-means clustering into a predetermined number k of clusters. The number k of clusters may then be used as a hyperparameter that may easily be optimized in only a few attempts.

[0029] Alternatively or in combination, at least one group of trajectory points may be formed by discretizing the space of the trajectory points into regular units and allocating the trajectory points in one particular regular unit to one particular group. For example, a two-dimensional space of trajectory points may be discretized into squares or rectangles in Cartesian coordinates, or wedges in polar coordinates. If a trajectory point then falls within one such square, rectangle or wedge, it may then be attributed to the group represented by this square, rectangle or wedge. In one example, such an assignment of trajectory points to groups may be used as a starting point for a later refinement. In three-dimensional space, squares and rectangles may become boxes, and wedges may become pie slices.

[0030] Alternatively or in combination, at least one group of trajectory points may be formed by setting up an arrangement of groups with free parameters, such that the assignment of trajectory points to these groups depends on the free parameters. These free parameters may then be optimized according to a predetermined optimization criterion. This optimization criterion is not limited to the assignment of trajectory points to groups as such. Rather, this optimization criterion may also relate, e.g., to and end result achieved after determining the final safety configuration, such as a smaller use of space on the factory floor, and / or a maximal operating speed of the robot.

[0031] In a particularly advantageous embodiment, the at least one robot is operated according to the determined safety configuration. In this manner, the advantages gained by the improved security configuration, such as a saving of space on the factory P240530W001 - 7 - 22.08.2024 floor and / or an improved operating speed (cycle time) of the robot, may be put into practice.

[0032] To this end, as discussed before, zones corresponding to groups of trajectory points, and corresponding velocity limits, may be loaded onto a safety controller for positiondependent enforcement of the respective velocity limits. Alternatively or in combination to this, at least one monitored safety zone may be established around the robot according to the determined safety configuration. This monitored safety zone may be monitored by any suitable means, such as light curtains or laser scanners. In particular, in response to a person reaching or otherwise entering the at least one monitored safety zone, motion of at least one robot may be slowed down or stopped.

[0033] Because it is computer-implemented, the present method may be embodied in the form of a software. The invention therefore also relates to a computer program with machine-readable instructions that, when executed by one or more computers and / or compute instances, cause the one or more computers and / or compute instances to perform the method described above. Examples for compute instances include virtual machines, containers or serverless execution environments in a cloud. The invention also relates to a machine-readable data carrier and / or a download product with the computer program. A download product is a digital product with the computer program that may, e.g., be sold in an online shop for immediate fulfilment and download to one or more computers. The invention also relates to one or more compute instances with the computer program, and / or with the machine-readable data carrier and / or download product.

[0034] DESCRIPTION OF THE FIGURES

[0035] In the following, the invention is described using Figures without any intention to limit the scope of the invention. The Figures show:

[0036] Figure 1: Exemplary embodiment of the method 100 for determining a safety configuration 2 for a robot workstation 1 ;

[0037] Figure 2: Exemplary division of trajectory points 6 of a trajectory 5 into groups 6a-6f (Figure 2a) and resulting monitored safety zone 2a (Figure 2b); P240530W001 - 8 - 22.08.2024

[0038] Figure 3: Example of rearranging the dividing into groups 6a-6f and reducing corresponding velocity limits 7a-7f in order to meet a constraint 2a* for the monitored safety zone 2a.

[0039] Figure 1 is a schematic flow chart of an embodiment of the method 100 for determining a safety configuration 2 for a robot workstation 1. The robot workstation 1 comprises at least one industrial robot 3. The safety configuration 2 that is sought comprises a combination of at least one monitored safety zone 2a and at least one velocity limit 7a- 7f of the robot 3 that are arranged such that, at the moment where reaching or other entry of a person into the monitored safety zone 2a is detected, there is sufficient time to stop the robot 3 before it can possibly come into contact with the entering person.

[0040] In step 110, locations 4a in space, as well as velocities 4b, of a least one reference point 4 of the robot 3 at multiple trajectory points 6 along a present or future trajectory 5 of the robot 3 are determined.

[0041] According to block 111, this determining 110 may comprise simulating a trajectory 5 of the at least one reference point 4 of the robot 3.

[0042] In step 120, the trajectory points 6 are divided into groups 6a-6f.

[0043] According to block 121 , at least one group 6a-6f of trajectory points 6 may be formed by unsupervised clustering.

[0044] According to block 121a, trajectory points 6 may be clustered into a predetermined number k of clusters by k-means clustering.

[0045] According to block 122, the space of trajectory points 6 may be discretized into regular units. According to block 123, the trajectory points 6 in one particular regular unit may then be allocated to one particular group 6a-6f.

[0046] According to block 124, an arrangement of groups 6a-6f with free parameters may be set up, such that the assignment of trajectory points 6 to these groups 6a-6f depends P240530W001 - 9 - 22.08.2024 on the free parameters. According to block 125, the free parameters may then be optimized according to a predetermined optimization criterion.

[0047] In step 130, for each group 6a-6f, a velocity limit 7a-7f is determined based at least in part on the maximum velocity of the reference point at any of the trajectory points 6 in this group 6a-6f.

[0048] In step 140, the safety configuration 2 is determined based at least in part on the locations 4a of the reference point 4 of the robot 3 at the trajectory points 6 on the one hand, and the velocity limits 7a-7f applicable to these trajectory points 6 on the other hand.

[0049] In particular, according to block 141, for each trajectory point 6, based on the velocity limit 7a-7f applicable to this trajectory point, a required separation distance 8 for persons from the location 4a of the reference point 4 of the robot 3 at this trajectory point 6 may be determined. According to block 142, the safety configuration 2 may then be determined based at least in part on these required separation distances 8.

[0050] According to block 142a, the determining 142 of the safety configuration 2 based at least in part on the required separation distances 8 may comprise determining at least one monitored safety zone 2a based at least in part on the required separation distances 8.

[0051] According to block 142b, the required separation distances 8, and / or an aggregate thereof, may be compared with at least one given monitored safety zone 2a, 2a*. In particular, a concrete given constraint 2a* may be imposed on the monitored safety zone 2a. It may then be checked in block 142c whether a required separation distance 8 for a trajectory point 6 exceeds the monitored safety zone 2a, e.g., by virtue of a then-necessary monitored safety zone 2a violating the given constraint 2a*. If this is the case (truth value 1), according to block 142d, the velocity limit 7a-7f for the group 6a-6f to which this trajectory point 6 belongs is reduced, thereby reducing the required separation distance 8.

[0052] In the example shown in Figure 1 , in step 150, the at least one robot 3 may be operated according to the determined safety configuration 2. P240530W001 - 10 - 22.08.2024

[0053] In step 160, at least one monitored safety zone 2a may be established around the robot 3 according to the determined safety configuration 2. It may then be checked in step 170 whether a person reaches or otherwise enters the at least one monitored safety zone 2a. If this is the case (truth value 1), in step 180, motion of the at least one robot 3 may be slowed down or stopped.

[0054] In step 190, zones 9a-9f in the space of the trajectory points 6 corresponding to the groups 6a-6f, as well as velocity limits 7a-7f corresponding to these groups 6a-6f, may be loaded onto a safety controller 3a of the robot 3. This safety controller 3a is configured to enforce these velocity limits 7a-7f independently of a motion controller 3b of the robot 3. The motion controller 3b may be the entity from which the trajectory 5 with its trajectory points 6 is obtained.

[0055] Figure 2a shows an exemplary situation at a robot workstation 1 where a reference point 4 of a robot 3 centered at a point P shall move along a trajectory 5 that leads from a point A to a point B and onwards to a point C before returning on a direct path back to point A. The robot 3 and its reference point 4 are not shown in Figure 2 for clarity. Individual trajectory points 6 of the trajectory 5 are labelled with their respective velocities that the reference point 4 has when passing there. The maximum reach R of the robot 3 is a circle around the point P.

[0056] In the example shown in Figure 2a, the trajectory points 6 are grouped into four groups 6a-6d, based mainly on location. Each group 6a-6d has many trajectory points 6 with similar velocities. The highest velocity occurring in each group 6a-6d is set as the velocity limit 7a-7f corresponding to the respective group 6a-6d.

[0057] Figure 2b shows the effect that the introduction of groups 6a-6d has on the monitored safety zone 2a. If only the globally highest velocity was to be considered, then the circle G around point P would have to be set as the monitored safety zone 2a. But in the present situation, where the highest velocity limit 7b of 3840 mm / s is only applicable to the group 6b and all other velocity limits 7a, 7c and 7d for the other groups 6a, 6c and 6d are lower, a smaller monitored safety zone 2a suffices to ensure that the velocity-dependent requirement for a safe separation between an entering human and the robot 3 is met at all times. P240530W001 - 11 - 22.08.2024

[0058] Figure 3a shows the same situation at the robot workstation 1 as Figure 2a. But in contrast to Figure 2a, the task is different: As it is evident from Figure 3b, there is a constrain 2a* in place that the monitored safety zone 2a is not to exceed a particular boundary y=1600 mm shown as a dashed line.

[0059] A comparison of Figure 3b with Figure 2b reveals that this requirement cannot be met with the velocity limits 7a-7d that have given rise to the monitored safety zone 2a shown in Figure 2b. At least some velocity limits have to be reduced. In order to reduce the total price of this in terms of cycle time, the division of the trajectory points 6 into groups is redone in a more fine-grained manner to yield new groups 6a-6f. As it is evident from Figure 3a, the advantage of this is that in groups 6a and 6b, the velocity limits 7a and 7b do not have to be reduced at all compared with Figure 2a. For group 6c, the velocity limit 7c has to be reduced only to 80 % of the velocity limit applicable in Figure 2a for the same area. Only groups 6d to 6f have to suffer from more drastic reductions of their respective velocity limits 7d to 7f to meet the constraint 2a* with respect to the monitored safety zone 2a.

[0060] P240530W001 - 12 - 22.08.2024

[0061] List of reference signs:

[0062] 1 robot workstation

[0063] 2 safety configuration for robot workstation 1

[0064] 2a monitored safety zone

[0065] 2a* constraint for monitored safety zone 2a

[0066] 3 robot at robot workstation 3

[0067] 3a safety controller of robot 3

[0068] 3b motion controller of robot 3

[0069] 4 reference point of robot 3

[0070] 4a location of reference point 4

[0071] 4b velocity of reference point 4

[0072] 5 trajectory

[0073] 6 trajectory points of trajectory 5

[0074] 6a-6f groups of trajectory points 6

[0075] 7a-7f velocity limits for groups 6a-6f

[0076] 8 required separation distances

[0077] 9a-9f zones corresponding to groups 6a-6f

[0078] 100 method for determining safety configuration 2

[0079] 110 determining locations 4a, velocities 4b

[0080] 111 simulating trajectory 5

[0081] 120 dividing trajectory points 6 into groups 6a-6f

[0082] 121 forming group 6a-6f by unsupervised clustering

[0083] 121a using k-means clustering

[0084] 122 discretizing space of trajectory points 6 into regular units

[0085] 123 allocating points 6 in unit to respective group 6a-6f

[0086] 124 setting up assignment of groups with free parameters 6a-6f

[0087] 125 optimizing free parameters

[0088] 130 determining velocity limits 7a-7f

[0089] 140 determining safety configuration 2

[0090] 141 determining required separation distances 8

[0091] 142 determining safety configuration based on separation distances 8

[0092] 142a determining monitored safety zone 2a based on separation distances 8

[0093] 142b comparing separation distances 8 to zone 2a, constraint 2a* P240530W001 - 13 - 22.08.2024

[0094] 142c checking whether zone 2a, constraint 2a* exceeded

[0095] 142d reducing velocity limits 7a-7f

[0096] 150 operating robot 3 according to safety configuration 2a

[0097] 160 establishing monitored safety zone 2a

[0098] 170 detecting that person reaches or enters safety zone 2a

[0099] 180 slowing down or stopping motion of robot 3

[0100] 190 loading zones 9a-9f, velocity limits 7a-7f onto safety controller 3a

[0101] A, B, C special points on trajectory 5

[0102] G monitored safety zone with only one global velocity limit

[0103] P center point

[0104] R maximum theoretical reach of robot 3 v velocity x, y coordinates in space

Claims

P240530W001 - 14 - 22.08.2024Claims:

1. A method (100) for determining a safety configuration (2) for a robot workstation (1) that comprises at least one industrial robot (3), wherein the safety configuration (2) comprises a combination of at least one monitored safety zone (2a) and at least one velocity limit (7a-7f) of the robot (3) that are arranged such that, at the moment where reaching or other entry of a person into the monitored safety zone (2a) is detected, there is sufficient time to stop the robot (3) before it can possibly come into contact with the entering person, the method (100) comprising the steps of:• determining (110) locations (4a) in space, as well as velocities (4b), of a least one reference point (4) of the robot (3) at multiple trajectory points (6) along a present or future trajectory (5) of the robot (3);• dividing (120) the trajectory points (6) into groups (6a-6f);• for each group (6a-6f), determining (130) a velocity limit (7a-7f) based at least in part on the maximum velocity of the reference point at any of the trajectory points (6) in this group (6a-6f); and• determining (140) the safety configuration (2) based at least in part on the locations (4a) of the reference point (4) of the robot (3) at the trajectory points (6) on the one hand, and the velocity limits (7a-7f) applicable to these trajectory points (6) on the other hand.

2. The method (100) of claim 1, wherein the determining (140) of the safety configuration (2) comprises:• determining (141), for each trajectory point (6), based on the velocity limit (7a-7f) applicable to this trajectory point, a required separation distance (8) for persons from the location (4a) of the reference point (4) of the robot (3) at this trajectory point (6); and• determining (142) the safety configuration (2) based at least in part on these required separation distances (8).

3. The method (100) of claim 2, wherein the determining (142) of the safety configuration (2) based at least in part on the required separation distances (8) comprises: determining (142a) at least one monitored safety zone (2a) based at least in part on the required separation distances (8).P240530W001 - 15 - 22.08.20244. The method (100) of any one of claims 2 to 3, wherein the determining of the safety configuration (2) based at least in part on the required separation distances (8) comprises:• comparing (142b) the required separation distances (8), and / or an aggregate thereof, with at least one given monitored safety zone (2a, 2a*); and• in response to determining (142c) that a required separation distance (8) for a trajectory point (6) exceeds the monitored safety zone (2a), reducing (142d) the velocity limit (7a-7f) for the group (6a-6f) to which this trajectory point (6) belongs, thereby reducing the required separation distance (8).

5. The method (100) of any one of claims 1 to 4, wherein at least one group (6a- 6f) of trajectory points (6) is formed (121) by unsupervised clustering.

6. The method (100) of claim 5, wherein trajectory points (6) are clustered (121a) into a predetermined number k of clusters by k-means clustering.

7. The method (100) of any one of claims 1 to 6, wherein at least one group (6a- 6f) of trajectory points (6) is formed by:• discretizing (122) the space of the trajectory points (6) into regular units; and• allocating (123) the trajectory points (6) in one particular regular unit to one particular group (6a-6f).

8. The method (100) of any one of claims 1 to 7, wherein at least one group (6a- 6f) of trajectory points (6a-6f) is formed by:• setting up (124) an arrangement of groups (6a-6f) with free parameters, such that the assignment of trajectory points (6) to these groups (6a-6f) depends on the free parameters; and• optimizing (125) the free parameters according to a predetermined optimization criterion.

9. The method (100) of any one of claims 1 to 8, wherein the determining (110) the locations (4a) in space and the velocities (4b) of the at least one reference point (4) of the robot (3) comprises simulating (111) a trajectory (5) of the at least one reference point (4) of the robot (3).

10. The method (100) of any one of claims 1 to 9, further comprising: operating (150) the at least one robot (3) according to the determined safety configuration (2).

11. The method (100) of any one of claims 1 to 10, further comprising: loading (190) zones (9a-9f) in the space of the trajectory points (6) corresponding to theP240530W001 - 16 - 22.08.2024 groups (6a-6f), as well as velocity limits (7a-7f) corresponding to these groups 6a-6f), onto a safety controller (3a) of the robot (3) that is configured to enforce these velocity limits (7a-7f) independently of a motion controller (3b) of the robot (3).

12. The method (100) of any one of claims 1 to 11 , further comprising: establishing (160) at least one monitored safety zone (2a) around the robot (3) according to the determined safety configuration (2).

13. The method (100) of claim 12, further comprising: in response to a person reaching or otherwise entering (170) the at least one monitored safety zone (2a), slowing down or stopping (180) motion of the at least one robot (3).

14. A computer program, comprising machine-readable instructions that, when executed by one or more computers and / or compute instances, cause the one or more computers and / or compute instances to perform the method (100) of any one of claims 1 to 13.

15. A non-transitory machine-readable data carrier and / or a download product with the computer program of claim 14.

16. One or more computers with the computer program of claim 14, and / or with the machine-readable data carrier and / or download product of claim 15.

Images